Abstract
In this paper, we propose a novel approach towards human action recognition using multi-resolution feature extraction. It is based on 2D Discrete Wavelet Transform (2D-DWT), where features are extracted from sequential video frames. The proposed feature selection algorithm offers an advantage of very low feature dimension and therefore, lower computational cost. The use of wavelet-based features enhances the distinguishability of different actions, which means it provides a very high within-class compactness as well as between-class separability of the extracted features. Moreover, certain undesirable phenomena (e.g., camera movement, variations in depth of the moving subject) are less severe in the frequency domain. Principal Component Analysis (PCA) is exploited to reduce the dimensionality of the feature space. We consider Euclidean distance-based classifier as well as Support Vector Machine (SVM) for classification. We do extensive experimentations on two benchmark action databases. It is found that t...
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have